212388
Chapter 5
Measured denitrification and nitrous oxide fluxes in intertidal mudflats
along the Zeeschelde (Scheldt estuary, Belgium)
Van Damme S., Starink M., Van derNat J., Struyf E., Van Cleemput O. & Meire p.
Abstract
Seasonal measurements were carried out
of denitrification, using the acetylene inhibition
method on laboratory incubated sediment samples, concurrent nitrate consumption rates, and
NzO emission measurements in field and laboratory conditions along a longitudinal and
vertical gradient of mudflats in the Scheldt estuary, Belgium. Additional experiments were
carried out
to
determine limitation
of
nitrate vs. carbon, oxygen consumption and to
investigate the role of flooding regime on NzO emission. Spatial and temporal variation
of
the gaseous emissions was high and correlation with ambient parameters was generally not
significant. Laboratory incubated denitrification values ranged from0.22 to 6.8 mmol N
d-1, concurrent N2O emissions ranged
m-2
from 0.005 to 0.48 mmol N m-2 d-l and freld incubated
Nzo emissions at low tide ranged from 0 to 0.49 mmol N
*-t d-t.
Although nitrate was
limiting for denitrification, N2O emission at 10kPa acetylene incubation and constant nitrate
availability decreased with rising
tide.
macrozoobenthos (oligochaeta) was
Oxygen consumption related with the presence of
with 1.23 mol 02 mol C-r d-r higher than previously
It is hypothesised that benthos affects denitrif,rcation not only by enhancing nitrate
transport from the overlying water into the sediment, but also by reducing the oxygen
assumed.
concentration at the water sediment interface.
93
Denitrification and nitrous oxide fluxes
5.1 Introduction
Among the numerous goods and services provided by estuaries, nutrient cycling is often
prominent. In one case (Shepherd et a\.,2007), nitrogen cycling amountedup to nearly the
complete estuarine habitat value as given by Costanza et al. (1997). Denitrification is the
main process that accounts for the ecosystem services of the nitrogen cycle, although
it
is
directly linked with emission of trace gas involved in global warming and stratospheric ozone
destruction (Seitzinger et al.,1984;1988). In many studies on N-cycling in estuarine habitat,
denitrification is determined in different indirect ways: as
a
rest fraction in mass balances (e.g.
Middelburg et al., 1995a; van Damme et a1.,2009) or isotope budgets (e.g. Gribsholt et al.,
2005), or through modeling (e.g. Soetaert et al., 1995; vanderborght et al., 2002). Direct
measurements
of
estuarine denitrification,
microsensors, direct measurement
measurements
with the
acetylene inhibition technique,
of N2 emission in a He background or
isotope
of nitrogen gas are less common. For instance, several ecological models,
including benthic and pelagic denitrification, are available ofthe Scheldt estuary (Soetaert et
al., 1995; Vanderborght et al., 2002). In contrast, direct denitrification measurements
are
restricted to one study and one sampling station (Laverman et a1.,2007). Such measurements
are more important than indirect determinations, as these could be more biased than
previously thought. Indeed, in recent years our understanding ofthe estuarine nitrogen cycle
has been amended with several new pathways, including anaerobic ammonium oxidation to
N2 (anammox). Annamox has a well documented importance
environments (Meyer et al., 2005), but its role
in
estuarine
in marine and coastal
N-cycling is still under
investigation. only a few estuaries have been examined so far, showing e.g. that
N2
production due to annamox ranged between 0 and22"/o in the Chesapeake Bay (Rich et al.,
2008). Annamox could bias nitrification coupled denitrification estimates such as presented
by Middelburg et al. (1995a). The acetylene inhibition technique, a well established method
to measure denitrification (Seitzinger et al.,1993), is unaffected by annamox as this process is
almost completely inhibited
at an
acetylene concentration
of 22 pM while
effective
denitrification inhibition to NzO requires 4mM (Jensen et aI.,2007).
The role of benthic denitrification in the Scheldt is gaining relative importance on pelagic
denitrification, as the ecological recovery ofthe estuary leads to higher oxygen concentrations
in the pelagic environment(Cox et a/., submitted); the Scheldt estuary is but one of
several
systems in this situation (e.g. Billen et a1.,2005).
95
Chapter 5
It
is not the aim to assess the influence of new pathways on previous denitrihcation estimates,
but to present data obtained by a reappraised method of a system that still lacks data. We
report the first measured denitrification data of mudflat sediments of the Scheldt estuary that
cover both the salinity gradient and a vertical gradient in the freshwater patI, as measured
with the acetylene inhibition technique. NzO emission rates were recorded additionally. A
comparison between the benefits of increasing benthic denitrification and the negative effect
of NzO emission is useful within the framework of ecosystem goods and seryices, as
the
greenhouse gas issue is of ever growing importance.
5.2 Material and methods
5.2.1 Study area
The Scheldt estuary is located in Northem Belgium (Flanders) and the Southwest Netherlands
(Fig.l). It
extends from the mouth at Vlissingen (km 0)
movement is stopped through a complex
till
Gent (km 158); there tidal
of sluices. The lower and middle estuary, the
Westerschelde (55 km long), situated in the Netherlands, is a well mixed region characterized
by a complex morphology with flood and ebb channels surrounding several large intertidal
mud and sand flats. The surface area of the Westerschelde is 310 km2, with the intertidal area
accounting for
35%o
of the area. The Sea Scheldt, situated in Belgium, is single channeled and
its surface amounts to only 44 km2. The water quality of the Scheldt has been heavily
impacted, especially in the Sea Scheldt (Van Damme et a1.,2005), but is now in aphase of
partial recovery (Soetaert et a1.,2006). This study is confined to the Sea Scheldt (105 km
long).
5.2.2 Sampling sites
Four intertidal mudflat sites were selected along the Sea Scheldt, largely on the basis of
salinity and accessibility (Fig. 5.1). Groot Buitenschoor (GB) is situated in the brackish part,
Burcht (BU) on the transition zone between the freshwater and the brackish zone, Durme
(DU) and Appels (AP) in the freshwater part (Table 5.1). In Appels a lateral gradient of four
stations was installed between low water level and the Scirpus spp. vegetation bordering the
marsh (Table 5.1). Wooden boardwalks were constructed and used to reduce the disturbance
caused by visits.
96
Denitrification and nitrous oxide fluxes
campaigns were conducted on a monthly basis during 1996 (N2o emission) and 1997 (Nzo
emission and denitrifi cation).
o4'20
04"40
?-ll|:Sfg: *,s.*l'*
Fig.5.1: Map of the selected sites
Table 5.1: Site description
Station
Elevation Moistue
TAW) (% DW)
(m
GB
BU
DU
API
APz
AP3
AP4
3.9
4.2
4.5
3.2
3.8
4.7
5.1
95
48
r 13
r 19
!
69 !
8l r
98 r
97 !
146
22
20
22
30
37
Conductivity
pH
C tot
(pS cm'r;
!
!
1504 r
1002 I
1028 I
ll38 r
lMz !
t5146
2517
9006
t536
319
362
281
206
344
(%wt)
I
I
t
7.5 r
7.5 r
7.3 r
7.5 !
7.8
8.0
7.8
0.3
3.0
0.1
r.2
O.2
3.6
0.1
1.2
0.1
1.6
0.1
2.5
0.2
4.5
i
!
I
r
!
0.5
0.2
0.8
0.6
0.2
1 0.8
t 1.2
NHo CaCO3 Sand fom
(o/owl) (%wt) (%) (0/.)
!
I
r
r
r
8.7 r
3.5 r
4.7
10.7
19.3
10.9
10.6
4.3 t7
5.0 12
10.8 9.9
IO.2 8.4
10.3 9.0
10.7 8.3
5.2 8.7
30
60
52
'15
67
48
l8
46
29
25
t4
l8
32
47
Clay
(%)
25
ll
23
t
t
15
2l
35
5.2.3 Field flux measurements
The release of nitrous oxide was measured by sampling accumulated gas beneath chambers
placed over the sediment. Before placing them -about 4 cm deep in the sediment- the circular
top lid was removed. This allowed chamber ventilation with ambient air while the sediment
could re-establish equilibrium after the possible sediment disturbance due to placement of the
chamber. The gas chambers had a circular ground surface of 176.7 cm2, a height of l0cm and
were made of non transparent polypropylene. The top lid contained a gastight septum at the
top that was used for connection with Teflon vacutainers by means of a double needle, so that
headspace samples could be transported and stored for analysis in the lab.
Samples were taken at t0 (i.e. immediately after closing the top lid) and half hour intervals
during one hour (1.e. 3 in total per replicate).
97
Chapter 5
5.2.4 Laboratory flux measurements
Sediment surface samples (-top 3 cm) were taken to the laboratory and homogenised by
gentle stirring. Subsamples of 40 g fresh weight were then put into preweighed airtight glass
recipients (volume 218.4
+ 0.8 ml, sediment horizontal
surface 15.8 cm2). For every
sampling, fluxes were determined simultaneously in 6 sets of subsamples, with following
treatments at two controlled temperatures (field temperature at moment
of sampling
and
250C):
.
.
NzO from sediment with a 40 ml layer of 0.66 mM (10 mg N L-r) KNO: solution
NzO from sediment with a 40 ml layer of 0.66 mM KNO: solution and with injection
of l0 kPa acetylene in
.
the headspace
Nitrate water-sediment flux from
a
40 ml laver of 0.66 mM KNO3 solution
As soon as the solution was added, benthic invertebrates emerged from the sediment
and
showed activity that persisted during the entire flux measurements. Nitrous oxide fluxes
under laboratory conditions (3 replicates) were determined by taking samples at t0 and at 24h
intervals during 2 days. At this time lag under steady state conditions, all fluxes, calculated
by regression analysis from the recorded change ofconcentration over time in the headspace,
were significant (p< 0.05). Acetylene (10 kPa) was injected into the headspace to block
conversion of NzO to N2 during denitrification. Shortcomings of the technique were thus
limited by applying the incubations on small sediment volumes (40 g), enhancing
penetration of the acetylene gas, and by providing long incubation times
the
(3d). Nitrate and
ammonium fluxes were determined by sampling the overlying water layer in the separate
subsamples at four to eight regular time intervals in separate replicates of a two days covering
time series.
5.2.5
N vs. G limitation experiment
To investigate whether N or C was a limiting factor for denitrifrcation, nitrate flux and NzOemission, laboratory flux measurements using homogenised sediment (as described above)
were conducted at four different concentrations of nitrate (0, 1,5 and l0 mg N L-r as KNO3)
at25"C. This experiment was carried out twice for sediment op Appels 2, Burcht and Groot
98
Denitrification and nitrous oxide fluxes
Buitenschoor, and 3 times for Durme. For each of these stations one experiment was carried
out with and without addition of I % glucose in the overlying solution.
5.2.6 Oxygen consumption experiment
In the same setup as the laboratory flux measurements (at 25"C), oxygen consumption from
a'wrw oxl 9l'oxygen meter. This was done on
(samplings of l0 and 2010911996), Burcht (ll and
the overlying water was measured with
sediments
of
Groot Buitenschoor
19/0911996), Durme (3 replicates on sampling
of ll/0111996 and 2 replicates on sampling of
Appels 2 (10 and 1810911996) and Appels 3 (1'1109/1996). Sediment of Appels
1610311996),
3, that contained almost no benthic macro-invertebrates, was amended with and without
addition
density
of
of
100 individual Oligochaetes, collected on a 250p sieve, corresponding with a
63290 ind. m-2 which is representative for an average site in the oligohaline part
of
the estuary (Seys el al.,1999).
Oxygen consumption was calculated by determining the slope between -l0o/o and -90% of the
total oxygen concentration
decrease during
the experiment, relative to the initial
concentration.
5.2.7 Experimental tidal mudflat
A mixture of (l:4) garden soil
and mudflat sediment (Burcht) was mixed and transferred to a
container(lxwxh:2.0 x0.5 x 0.8 m).
The containerwasplacedin atemperature
(18+0.5'C), humidity (70+7%) and CO2 concentration (380+40 ppmv) controlled room. The
sediment was allowed to settle for a period of 4 months. After this period sediment height
was
-35 cm. Two polypropylene collars with
an inner diameter of
2l
cm were installed in the
centre of the container to ascertain consistent placement of the gas collecting chambers and
minimise disturbance during successive emission measurements (Van der Nat & Middelburg,
2000). The flux chambers were 80 cm high (volume 55.4
l). A 16 hours day period with a
light intensity of -0.4mmol photons m-2.s-l at the sediment surface (about 1.6 m below lamps)
was set, except during the measurements when light was continuously supplied to avoid
interference of the light regime with N2O emission. Two tidal regimes (6 hours low
high tide, and l0 hours low
-
-
6 hours
2 hours high tide resp.) were installed, each one during one
week of which three days were used for allowing the sediment to come into equilibrium with
the imposed tidal regime. The tidal range was -16 cm. A large basin (-500 l) was used as a
99
Chapter 5
reservoir for the overlying water. The nitrate concentration of the overlying water was daily
adjusted at l0 mg N L-'.
5.2.8 Analysis
Gas samples were transferred from the vacutainers or laboratory recipients to the detection
equipment by means
of 'Hamilton-GASTIGHT@' syringes. NzO was
determined using a
Chrompack@ 437 A gas chromatograph, equipped with a stainless steel Altech Chromosorb
102 column of 4.88 m length ancl 3.175 mm diameter and a
63Ni
electron capture detector,
under the following conditions: injector temperature 90oC, oven temperature 90"C and
detector temperahtre 300'C. Standard gas of 51.4 ppmv NzO in helium was used for
calibration over a range of 10
-
1000 pL injection volume. For high tide conditions the NzO
concentration of the water phase was taken into account according to Moraghan
&
Buresh
(1977).Nitrate and ammonium in water were determined according to Bremner (1965a).
Total carbon of the sediment was determined on air dry sediment according the method of
Walkley
&
Black (Allison, 1965). Nitrate and ammonium in sediment were determined
according Bremner (1965a) on extraction of 40 g sediment, extracted during
3N
KCl. Total N was determined after Kjeldahl
Sediment temperature was measured
lh with 30 ml
destruction according Bremner (1965b).
with a sediment
thermometer
at 2 cm
depth.
Conductivity of the pore water was measured with a WTW LF 91' conductivity-meter. pH of
the sediment was measured in a sediment suspension according to Verloo (1988) with aC832
Consort. Sediment texture was determined with a Sedigraph 5100. The CaCO3 content of
sediment was determined throush titration with HCl.
Estimates for annual nitrous oxide emission rates were calculated for all sites by integration
of
the curves connecting averages of replicate measurements. Months without data were
interpolated
linearly. QlO values were
determined on pair wise incubations
at field
temperature and25 "C, and determined only if this temperature difference was at least 10"C.
5.3 Results
5.3.1 Site characterisation
The annual average sediment conductivity, carbon and nitrogen contents are listed in Table
5.1. The ammonium content of the sediment was in the freshwater stations generally higher
100
Denitrification and nihous oxide fluxes
than in the brackish stations of Burcht and Groot Buitenschoor. The vertical gradient in
Appels was generally characterised by an increasing total carbon content, clay and loam
fraction and moisture content in upward direction. Dissolved nitrate concentrations (data not
shown) were an order of magnitude lower than ammonium concentrations.
5.3.2 N2O-Emission rates
Emissions of nitrous oxide from intertidal sediments of the Schelde estuary, as sarnpled, in situ
at low tide, and those determined after addition of nitrate solutions in the laboratory were
highly variable, both temporal and spatial (Fig. 5.2-5.3). Negative fluxes were not recorded;
all fluxes were from the sediment into the atmosphere. Laboratory incubated denitrihcation
ranged from 0.22 to 6.8 mmol
N
m-2 d-l (Fig. 5.2), concurrent
Nzo emissions ranged from
0.005 to 0.48 mmol N m-2 d-r (Fig. 5.3a) and field incubated NzO emissions at low tide ranged
from 0 to 0.49 mmol N
m' d I 1Fig. 5.3b).
N2O emissions in the field and in the laboratory
after addition of nitrate solution thus were situated within the same range. On average, when
a l0 mg N L-r nitrate solution was applied at field temperature, NzO emission with addition
ace$rlene was on average
5l
of
times higher (3 to 345 times) than without acetylene. These high
values indicate that denitrification was effective under ample nitrate provision.
Denitrification removed between
3 and. 60%o (average 2l%) of the nitrate in
the overlying
water.
1
0000
9000
8000
7000
c
J
o
5000
o
4000
E
z
--l-""
6000
---I- BU
--*-DU
-.X_
l
AP1
r*AP2]
,-+-
3000
AP3i
+AP4
2000
I
1000
0
J FMAMJ
JA
SOND
Fig' 5.2: Sediment denitrification; incubation in laboratory at field temperature, after addition of 10 mg N
L-l nitrate solution
101
Chapter 5
1000
900
800
700
E
!t
q.E
j
o
E
o
z
600
--+-GB
500
--r--BU
400
:
--+-DU
-)<- AP1
300
----+-AP2
200
-+-
l
l
i
]
l
AP3r
,;;1AP41
100
0
-100
J FMAMJ
a)
1000
900
800
700
'itl
'E
o
600
500
E
400
o
z
300
200
100
0
-100
b)
Fig. 3: N2O emission from sediment; a) incubation in laboratory at field temperlture, after addition of 10
mg N L-r nitrate solution (deta from 1996); b) iz sitz incubation at low tide (data from 1997)
102
Denitrification and nitrous oxide fluxes
In general, maximal values were recorded during summer and minima during winter (Fig. 5.2-
5.3). Univariate analysis of variance on all replicates of the different sampling stations, with
sampling month as a sorting variable, showed for denitrification significant variances
(p<0.05) between sampling campaigns for all stations except Appels
5.2). For Appels 3, however, the variance was related with
I and Appels 2 (Table
a maximum denitrification value
in winter (Fig. 5.2). In Appels, 3 of 4 stations thus showed no seasonality for denitrification.
For laboratory incubated N2O emissions, variances between sampling months were significant
for all stations except for Burcht and Appels 2 (Table 5.2), but for Appels
I
and
4
the
significant differences were very small. For field incubated NzO emissions, not one variance
between sampling months was significant.
It can be concluded that the variance of the
data
was so large that seasonality was not clearly apparent.
Table 5.2: Univariate analysis of variance on all replicates of the different sampling stations, with
sampfing month as a sorting variable, for in situ and laboratory incubated N2O emission and
denitrification
Nzo
Nzo
in situ
df
BU
DU
API
AP2
AP3
AP4
Nzo
NO,--\ solution)
(10 mg NO,--\ solution)
(10 kPa acetylene)
dfFp
F
t2
t2
12
18
18
18
l8
GB
(10 mg
3.1 0.62
2.2 0.12
0.7 0.3s
1.6 o.il
0.2 0.23
3.9 0.64
0.6 0.54
12
12
t2
18
18
18
t8
10.4
2.6
4.5
6.5
0.3
111
9.3
0.004
0.1,2
0.04
0.004
0.9r
0.000
0.001
12
t2
12
l8
18
I8
I8
8.8
5.9
20.2
2.0
2.4
15.4
10.8
0.006
0.02
0.00
0.15
0.09
0.000
0.000
For laboratory incubated N2O emissions (Fig. 5.4a) and denitrification (Fig. 5.4b), variability
on the data was larger in the freshwater zone than at higher conductivity. This statement is,
however, biased by the higher number of samples in the freshwater zone when taking also
into account the data of the vertical gradient of Appels, as was done in Fig.5.4. Univariate
analysis of variance on data of only Durme, Burcht and Groot Buitenschoor, which are
distributed more uniformly along the salinity gradient, showed, with specific conductivity as a
sofiing variable, a signif,rcant effect of specific conductivity between Durme and Burcht
(denitrification:
F1,36
:
8.0, p
:
0.001; NzO emission: Fr,:o : 25, p
:0.004), but not between
Burcht and Groot Buitenschoor.
t03
Chapter 5
1000
900
800
700
600
o 500
E
400
o
300
6a
c
z
200
100
0
10000
15000
20000
Sp. Cond. (pS.cml)
10000
9000
8000
E
q 7000
E
6000
J
o 5000
E
4000
o
3000
N
2000
1000
z
0
15000
Sp. Cond. (pS.cmr)
Fig. 5.4: Sediment N2O emission (a) and denitrification (b) vs. specific conductivity of sediment interstitial
I
water; incubation in laboratory at 25oC, after addition of 10 mg N L nitrate solution
Univariate analysis of variance on all replicates, with location as a sorting variable, of the
different sampling campaigns of the stations in Appels showed zero significant (p<0.05)
differences for denitrification (data not shown). As such, for denitrification, no vertical
gradient was observed at all.
For laboratory incubated NzO emissions Appels
I
and Appels 4 showed both significantly
lower values than both Appels 2 and Appels 3 (API-AP2'. Fr;rs:6.3, p:0.018; API-AP3:
Fr,ra:35, p:0.009; AP4-AP2: Fr,re:2.3, p:0.002; AP4-AP3: FrJF42, p:0.012).
All
other
differences were not significant (data not shown).
For field incubated NzO emissions, only one station showed consistently lower
values.
Appels 4 showed significantly lower values than all other stations (APl: F1,a6:45, p:0.008;
AP2:Fr.qo:12, p:0.013; AP3: Fr,ao:l,8, p:0.024). No other significant differences between
stations were observed for field incubated NzO emissions (data not shown).
104
Denitrification and nitrous oxide fluxes
Careful examination of the data revealed not a single factor, or any combination of measured
variables, that could explain the temporal or spatial variation of the observed fluxes.
Estimates for annual nitrous oxide emission rates were calculated for all sites by integration
under the curves that connect monthly averages
of
replicate measurements (Table 5.3).
Months without data were interpolated linearly. QlO values ranged between 0.7 and. 2.6,
showing thus high variability but within the range of biological processes.
Table 5,3: Annual integrated N2O emission rates and Qr6 values
Station
GB
BU
DU
APl
AP2
AP3
AP4
5.3.3
NuO
NzO
in situ
(10 mg NOr--l\ solution)
(mmol rn2y-l)
(mmol m-2y-l)
+
+
+
8.1 + 4.7
18 * 15
84 +76
7.6 + 3.4
19 + 5.0
3l + 21
4.3 + 1.5
3.6
7;7
8.9
29
1.8
6.7
4.1
azJ
41
+48
26
+20
11
+
5.5
Nzo
QlO
Q10
(10 mg NO,--11 solution)
(l 0 kPa acetylene)
(mmol-_)-l
m
+ 1.2
+ 0.5
1.2 + 0.6
2.6 + 1.7
0.7 + 9.3
1.0 + 0.4
0.8 + 0.3
-y
1.5
0.9
314 +
1335 +
462 +
472 +
503 +
538 +
)
122 2.0 +
232 1.0 +
911 1.0 +
116 1.3 +
167 l.l +
154 t.3 +
3ll 0.9 +
1.4
0.'7
0.4
0.3
0.5
0.3
0.3
N vs. G limitation experiment
As the nitrate concentration increased in the overlying solution (0,
l,
5 and
l0 mg N L-]), both
the nitrate consumption rate and the denitrification rate increased (Fig. 5.5). Only in one case
the increase was not signihcant (nitrate water-sediment flux Burcht, slope 0.02, R2 0.01,
F1.3:8.2, p>0.5). Apart from this case the slope of nitrate water-sediment flux ranged from
0.32 to 2.14 (R2 ranging from 0.78 to 0.99), and the slope of denitrification ranged from 0.09
to 0.98 (R2 ranging from 0.82 to 0.99). Nitrate disappeared 2 to 5 times faster from
the
overlying solution than the according denitrification increase (Fig. 5.5). An additional similar
experiment where the sediment was added in different recipients so that the water-sediment
surface was 12.6 vs. 315 crn2, i.e.25 times larger, ceteris paribus, showed that the time before
the overlying nitrate solution dropped from 10 to 2 mg N L-r, was 4 days resp. 4hours, i.e.24
times shorter. The concordance between the surface increase and the nitrate consumption rate
105
Chapter 5
decrease indicates that the reactive surface, hence the water sediment transport, was limiting
for nitrate removal of the overlying solution.
0,8
c
o
(g
(J
0,6
'tr
=c
Ito
0,4
o
CL
o
o
0,5
-0,5
1
1,5
2,5
Slope nitrate water-sediment flux
Fig. 5,5: Slope of denitrification vs. nitrate water sediment flux at increasing overlying nitrate solutions (0'
1, 5 and 10 rng N L-1); replicates over the
Slopes were determined on
nitrate concentration gradientl Dpg:31 rrtpz:2, 11nu:2, nce:2.
individual regression lines of the flux rates (in mmol m-2 d-t) at increasing
nitrate concentrations and then averaged over the different experiments.
Emission of nitrous dioxide did not show a similar increase as in Fig. 5.5 (data not shown);
slopes ranged from -0.03 to 0.
l6
1R2
from 0.32 to 0.77).
Amendment with glucose on replicate samples (addition
of l0 mg N L-' only) of the N vs.
C
limitation experiment did not lead to any change in nitrate water-sediment flux rates after the
hrst day of incubation (paired T-test, performed on individual nitrate water-sediment flux rate
slopes with and without glu amendment for the stations Durme, Appels 2, Burcht and Groot
Buitenschoor, t-0.86, p:0.89,
n:4).
These results show that nitrate availabiliry, not carbon,
was the limiting factor for denitrification in the Scheldt sediments.
5.3.4 Oxygen consumption experiment
Oxygen consumption was highest in sediment of the Durme and lowest in sediment of Appels
3 (Table 5.4). Addition of 100 individual oligochaetes on sediment of AP3 (experimental
sediment
106
surface: 15.8 cm2) resulted in a change from zero order to a polynomial
decrease
Denitrification and nitrous oxide fluxes
(Fig. 5.6). Given an oxygen consumption rate increase from 0.48 to 8.33 pg O2.cm-2 h-l
(Table 5.4), the oxygen consumption of one average oligochaete amounted to 1.2 pgoz h-t.,
corresponding with a consumption rate of 1.23 mol O2.mol C-l d l.
Table 5.4: Oxygen removal rates of mudflat sediment
Rate
Station
n
(mg orcm-2h-r)
GB
BU
1.3
2.3
7.9
2.0
DU
AP2
AP3
AP3+100 Oligochaetes
+
+
*
+
0.5
1.2
2
z
1.8
5
1.0
0.5
z
I
8.3
I
7
Y
+
6
C"
E b
tr
o
CD
4
J
+
AP3
.
AP3+100 Olig@haetes
x
o 3
AP3 + .t00 Otigo
It
_
Trend AP3
-Trend
----TrendBlank
o
.> 2
o
o
.9,
o
e|ant
1
0
120
180
Time (min)
Fig. 5.6: Dissolved oxygen concentration in overlying water of sediment with and without addition of 100
Oligochaetes;
blank:
water without sediment or Oligochaetes
5.3.5 Experimental tidal mudflat
Denitrification in sediment, exposed after submersion by l0 mg.L-rnitrate solution, and under
l0
kPa acetylene incubation, decreased exponentially with time (Fig.
low tide regime of 2h
- l0
5.7). At a high tide -
h, the measured denitrification rate decreased to zero after 8 to 9
hours, whereas at the 6h-6h regime, the slope of the exponential curve did not reach zero at
the end of the low
tide. This
pattern was repeated over 3 tides (data not shown). Organic
r07
Chapter 5
matter was not the limiting factor in the decrease, as the NzO concentrations in the head space
continued to increase after consequent low tide phases (Fig. 5.7).
30
25
-**
_
?20
E
CL
9rs
Exposed 6h
submerged 6h
Exposed 10h
o
210
Submerged 2h
0
8 10 12 '14 16 18 20 22
24
Tin€ (hours)
Fig. 5.7: Comparison ofN2O concentrations in the flux chamber ofthe tidal experiment container under
two tidal regimes: 6 hours low tide - 6 hours submersion with l0 mg Lr NOr--N, and 10 hours low tide - 2
hours submersion with
allowed to adapt
for
l0
mg L-t NO3--N. Before the measurements were recorded the meso-system was
1 day to the imposed conditions. The experiments were carried out under 10 kPa
acetylene incubation. The arrow indicates the start of the 10 hours low tide - 2 hours submersion run.
5.4 Discussion
Nitrous oxide emission rates were, with or without addition of acetylene in the flux chambers,
highly variable, both temporally and spatially, confirming in this aspect Middelburg et
a/.
(1996). Seasonality was only partly apparent although the Qro values indicated a temperature
effect (Table
5.2). The temporal or
spatial variation
of the emission data could not be
explained by any combination of variables.
5.4.1 Denitrification
The observed denitrification rates (Table 5.2) were for the conesponding locations (Durme
and Burcht) an order
of
magnitude lower than the denitrification rates estimated by
Middelburg et al. (1995b). Middelburg et
denitrification
at low tide, while in our
submerged sediment. According
sediments
is in most
al. (op cit.)
calculated nitrification coupled
approach acetylene inhibited nitrification in
to a review of Seitzinger et al. (1988), nitrification in
estuaries the major source
of nitrate for
sediment denitrification;
denitrification resulting from overlying nitrate being about ten times lower than nitrate
108
Denitrification and nitrous oxide fluxes
coupled denitrification. This is confimed by our denitrification results: they are one order
of
magnitude lower than the nitrification supported denitrification values of Middelburg et at.
(1995b), while the agreement with the diffusion supported {iaction is good: 0.50
mol
m-2 y-r
for Doel which is comparable with our values of the nearby station of Groot Buitenschoor.
There was also agreement with the modelled results of Soetaert & Herman (1995).
Laverman
et al.
(2007), who combined the acetylene technique with microsensor
of Appels, found depth integrated denitrification values of
measurements on sediment
mmol N m-2 d-l which is much more than our values of 0.22 to 6.8 mmol N m-2
d-r.
12
These
values, however, represent potential rates, stimulated by flow through conditions (Laverman
et al., 2006i). High denitrification values (up to 13.8 mmol N
intertidal freshwater sediments
technique (Garcia-Ruiz et
*-'
d-') were recorded for
of the Yorkshire Ouse, using the acetylene
al., 1998). The global
inhibition
averaged system denitrification values
(adding up both nitrification and overlying nitrate as sources) of estuaries (about I mmol N m2
d-r, excluding fresh water zones) and rivers (about 2 mmol N m-' d-'), as reviewed by
Seitzinger et al. (2006), indicate that denitrification in the Scheldt, according to any author, is
intensive.
The nitrate load leaving the Zeeschelde amounted in the early 1990's to 9000 tons per year
(Soetaert
& Herman, 1995).
Based on our data, extrapolated linearly for the pelagic nitrate
concentrations (as recorded by van Damme et aL.,2005) by using the slopes of Fig. 5.5, and
by using the compartments proposed by Soetaert & Herman (op cit.), the mudflats of the Sea
Scheldt could only eliminate 0.6% of this load by diffusion supported denitrification. Despite
the intensify of the process, diffusion supported denitrification in mudflats constituted
a
marginal effect on the nitrogen budget of the estuarine system. Including also nitrification
supported denitrification, Middelburg et al. (1995b) estimated that intertidal sediments might
account
for
l4%o
of the total estuarine nitrogen retention, making abstraction of
possible
bias. The recent evolution of the water quality (Cox et al., 2009), indicates that
despite the recovery of the oxygen status, the nitrate concentration continues to decrease so
annamox
that diffusion supported denitrification
will
decrease with
it,
as
nitrate was the limiting factor
for denitrification. At constant nitrate availability, however, the N2O emission rate decreased
as the inundation period and concurrent tidal submersion level increased (Fig. 5.7), although
the headspace concentration continued to increase. The depth profiles of Laverman et al.
(2007) show similar patterns at some mm in depth, as the pore water concentration increased
with inundation time, but
at the surface the production rate
did not significantly change.
109
Chapter 5
The stimulating influence of benthos, and more specifically of Oligochaetes, on
denitrification has already been reported (Chatarpaul et al., 1980; Pelegri and Blackbum
1995). This was explained by enhanced water-sediment transport. The peaking values of
denitrification in the sediment of the Durme (Fig. 5.2) were registered when this station was
characterised by a massive presence
of oligochaetes. On this location, Seys el al. (1999)
found an average benthos density of 243400 ind. m-2 which was about 4 times more than the
average density in the entire Zeeschelde. It was noted that the abundant benthos community
in the Durme site was removed by local dredging after the month of August, which could
explain the lower denitrification values in the Durme in October (Fig. 5.2). In order to further
assess the impact
of the benthos, comparative oxygen consumption rates were measured. The
sediment oxygen consumption increase after addition
of
100 individual oligochaetes was
of
the same range as the difference between the oxygen consumption rates of the Durme and the
other stations (Table 5.4). The benthos oxygen consumption recorded in this study (1.23 mol
02 mol C-t d-'), was more than 3 times larger than given by the benthos respiration formula at
25"C in the model of Soetaert et al. (1995). This indicates that the enhanced denitrification
rates
of the Durme station are probably linked with benthos activity. not only through
enhanced exchange
of nitrogen, but also because the benthic activity lowered the oxygen
content of the sediment surface layer.
5.4.2 Globalwarming
The in sita NzO emission rates at low tide (Table 5.2) were up to 4 times higher than the
corresponding values found by Middelburg et al. (1995b): as indicated in their paper, their
values could have been higher if they had applied longer accumulation times in the incubation
chamber.
The emissions of NzO were compared with other sources of greenhouse gas emissions in the
Schelde. The global warming potential of COz, CH+ and NzO (ratio 1:23:296), was multiplied
with the according annual emission rates, in order to rank the importance of the emissions per
surface unit according their contribution to global warming (Fig.
5.8). This comparison
revealed that COz is by far the most important contributor the greenhouse problem when
compared with CHa and NzO. Only in the freshwater tidal flats, enornous CIL emissions that
were recorded by Middelburg et al. (1996) in the Durme tributary caused CH+ to be the
dominant greenhouse gas (18 times the reference). These high values were not reported
110
Denitrification and nitrous oxide fluxes
elsewhere in the freshwater part (Siebens, 1997). The peculiar aspect of the Durme station, as
described earlier, is likely to be not representative for the freshwater part of the Schelde.
18.2
aco2
o CH4
tr N2O
r N2O this workr
FBS
FB
FBS
Channel
Mudflat LT
Mudflat HT
Fig. 5.8: Relative contribution of greenhouse gas emission fluxes of the Schelde to global warming
potential per surface unit for different compartments of the Scheldt esfuary, compared to the CO2 GWp
ofthe brackish channel compartment, set as reference value l, as measured during the period 1993-1997
(channsl = pelagic + subtidal sediment);
F:
salinity 0-2,
B:
salinity 2-15, S = salinity > 15, LT
:
low tide,
HT = high tide. Sources: CO2 channel: Frankignoulle e/ aL (1996); CIIa channel: Middelburg el a L (2002\,
fluxes calculated with piston velocity of 8.4 cm.h-r as determined by tr'rankignoulle et aI (f996); NrO
& de Bie (2000); CO, and CIIr mudflat: Middelburg et aL (lW6'); N2O mudflat:
Middelburg el aI (1995b) and this work, For Mudflat HT only data of this work (N2O) are available.
channel: De Wilde
Nitrous oxide emission from the water surface contributed 5 (fresh water part) to 13 times
(brackish and saline part) more to global warming than emission of CHa. In mudflats at low
tide, on the contrary, emission of NzO contributed marginally in comparison to COz and CH+.
For mudflats at high tide, no COz or CH+ emission data were available to compare with the
higher NzO emissions. Our data indicated that emission of N2O from exposed tidal flats were
an order of magnitude lower than emission from the water surface according to De Wilde
&
lll
Chapter 5
de Bie (2000). As the mudflat surface becomes submerged, the emission values shifted more
to those of the water surface above the subtidal system compartment.
The average NzO emission from cultivated land
76.4 mmol
m-'y-' for Belgium
is
101 mmol m-'
and Luxemburg (Boeckx
y-' for
the Netherlands and
& Van Cleemput, 2001). Our data
(Table 5.2) are, except for the peaking values of the Durme, lower than the agricultural
emissions around the Scheldt. As far as NzO is concemed, changing cultivated land into
mudflats results in a decrease of emission. Middelburg et al. (1995a) already pointed to the
marginal importance of estuarine NzO emissions on world scale. However, cultivated land is
a sink for CHa (Boeckx & Van
Cleemput, 2001), whereas mudflats are clearly not
(Middelburg et al., 1996). But it is possible that the estuarine greenhouse gas emission rates
that were observed during the nineties are changing considerably.
During recent years a regime shift took place in the Schelde (Cox et al., 2009). The system
has shifted from heterotrophy to autotrophy, as primary production has increased drastically.
While oxygen concentrations in the nineties were generally low, especially in the freshwater
zone (Van Damme et aI.,2005), nowadays supersaturation of oxygen is regularly observed
during summer, and ammonium concentrations have dropped to almost zero (Maris, pers.
com.). The question arises if the ratio between the different greenhouse gas emissions of the
estuarine system
will alter significantly during such changes. It is hypothesised that, since
the
nineties, emission of COz and CHa has shifted from the estuarine system to water treatment
plants, offering technical possibilities to capture CHa.
If
the hypothesis is true
it
can be
concluded that restoring estuarine ecological habitats by creating more intertidal areas cannot
be compromised by pointing to the medal backside of enhancing greenhouse gas emission.
Acknowledgements
This study was funded within the OMES project by the Flemish Environmental Agency
(VMM). We thank
the Flemish Administration for Waterways and Maritime Affairs, division
Zeeschelde for the held support.
112